Sailors awaiting rain aim straight, said Captain Craig.
Does that sentence make any sense to you? Me neither. It resulted when AI was provided with text and tasked to create something readable, but was misinterpreted by the app to mean that each word had to include “ai.” Whoops.
The lead sentence is, of course, laughable, but it’s certainly no joke if/when AI screw-ups cost real estate pros hard-earned commissions, clients and reputations. AI has certainly established the ability to save agents gobs of time by automating repetitive, time-consuming tasks. AI is great at supporting agents but terrible at trying to be one. Here are 12 things it cannot do, according to top real estate professionals around the country.
AI can’t read a room
“AI is an incredibly powerful tool, but where it consistently falls short in real estate is the judgment, nuance and accountability,” says Josh Jarboe, with REMAX Empire Buyers in Kentucky. “It can analyze data, draft content and speed up workflows, but it cannot replace human interaction or experience in the trenches. AI doesn’t read a room during negotiations, sense hesitation from a seller or recognize when a deal is about to fall apart because of emotion, ego or fear rather than the reality of the situation. Humans have instincts, and anytime our food, water or shelter feel threatened, primal emotions can often be triggered, which AI will never be able to understand.”
AI can’t see inside a house
“AI has not been in the houses we list to explain layouts and features, like examining the depth of a closet (will a standard hanger fit?), smells and ceiling height,” says Pam Rosser Thistle, with Berkshire Hathaway HomeServices Fox & Roach, REALTORS® in Philadelphia. “Also, an agent is aware of what’s to come. In Philadelphia, some properties have a tax abatement. An agent will know the terms to be able to advise the buyer about steep increases when the abatement expires. We are humans with both real estate and real life experience, and a mastery of our markets.”
AI can’t be hyperlocal
“When it comes to clients, they have to realize, especially in smaller metro areas, that the information isn’t as local as they like to think it is,” says Jeffrey Decatur, with REMAX Capital in upstate New York. “AI frequently pulls national headlines and makes sweeping generalizations in its answers. It doesn’t take into account that real estate is hyperlocal. Real estate laws and standards of practice differ from state to state, and sometimes from region to region. There are very limited situations where one answer covers the entire country. For example, in Albany, New York, I hired an AI company to create content for me, but it pulled info from New York City. AI has never viewed properties. It doesn’t know the nuances and differences of an area. AI can only take what information it is supplied with. It won’t live it or experience real estate the way an agent or client does.”
AI can’t know crucial details
“There are some things that AI consistently gets wrong in real estate,” says Monisha Rana, with Coldwell Banker Warburg, in New York City. “It has yet to accurately determine whether buildings are pre-war or post-war, for instance. In NYC, a building’s pre- or post-war status is often a key architectural and style distinction, and many people feel strongly about their preferences. On several occasions, when trying to find comps for a particular unit, I have input data into ChatGPT and asked it to give me similar buildings in the neighborhood, making sure to include whether the building was pre-war vs post-war, had a doorman, and any other amenities that would help me get as similar a comp as possible without being in the same building. On all of the occasions AI was very good at finding buildings in the neighborhood. However, it has not been accurate about whether the building was in fact pre-war or post-war, and moreover, AI has not been accurate or helpful in distinguishing which side streets or crossings are considered more desirable, such as less busy traffic-wise, have a school next door, etc.”
AI can’t build trust
“AI can analyze data, but it cannot replace human judgment,” says Madison Blau, with IS LUXURY in Las Vegas and California. “It cannot read emotional cues, body language or the subtle signals that matter most in negotiations, where timing, tone and intuition often outweigh data-driven responses. AI cannot build trust in a high-stakes, emotional transaction where clients need reassurance, empathy and accountability. Trust is built through consistency, transparency and lived experience, not algorithms. While AI can pull generalized data, it cannot truly understand micro-markets, street-by-street value differences or neighborhood culture, which come from being present and engaged locally.”
AI can’t communicate directly
“AI cannot replace the fundamental human connection we share with our clients,” says Robert Little, with REMAX Advantage in Henderson, Nevada. “While it can draft a message, it should be used sparingly in direct communication. Clients often sense when a response is automated, which can erode trust. Ultimately, AI cannot pick up the phone to offer reassurance during an emotional purchase or provide the sage, experience-based advice required during a difficult sale where empathy is essential.”
AI can’t create logical comps
“AI falls short in comparative analysis because it isn’t able to differentiate between renovations that sellers have made to a home or make adjustments for them,” says William Yau, with Coldwell Banker Warburg in New York City. “AI cannot understand development rezoning, natural light, air rights, etc., to provide the most accurate market value analysis. For example, there were two single-family homes near a construction area. One of them faced the construction site, which will soon impact its natural light (previously unobstructed; will now face a brick wall). The other home was a few blocks away and had more wear and tear. The comp analysis conveyed that the home facing the construction was valued higher. However, the home that sold for the higher price was the one farther from the construction zone despite its less desirable condition.”
AI can’t lead the path to closings
“Once a buyer is writing an offer, the real work begins,” says Eric Lavey, with The Beverly Hills Estates in California. “The transaction must be held together throughout escrow, knowing there is a 30%-plus chance it will never reach closing. Most failures are not about price or terms. They come from misunderstandings, an inability to recognize what a buyer or seller is feeling and the failure to bridge that gap toward clarity, comfort and trust. And this is only the beginning of the deal. Marketing a home or placing it on the MLS is not what a seller is truly paying for when they hire a broker. They are paying for judgment, perception and execution. Not every agent can do this, let alone AI.”
AI can’t perceive emotions
“AI is a great tool that processes and mirrors the information it is given,” says Shannon Murree, with REMAX Hallmark Chay Realty Brokerage in Barrie, Ontario. “It outputs only what it is instructed, with no judgment, intuition, empathy or awareness of what is missing. It cannot hear hesitation in someone’s voice, or sit with a client who is navigating grief, divorce or a major life shift and simply listen with intention. It cannot walk a property and notice what experience tells you matters: the smell, the sound, the feel, the things that never show up in data.”
AI can’t predict ahead
“Our team has integrated AI into nearly every aspect of our business but it’s not replacing our primary roles,” says Ivan Chorney, with Compass’s Ivan & Mike Team in South Florida. “There’s a local intelligence that never shows up in data or AI. Knowing that a developer quietly controls multiple homes on a street, that infrastructure is planned nearby or that a neighboring site may be headed toward demolition and years of construction. Those details can dramatically change how you value a property and whether you buy it at all. The best outcomes will come from professionals who use AI to enhance their judgment, not from AI trying to replace it.”
Like agents, AI can’t give legal/financial advice
“Whether putting together a contract or reviewing a building document, a small oversight can create a big problem,” says Blake Zises, with Coldwell Banker Warburg in New York City. “AI (like agents) should not give financial advice. When a seller reviews offers, they should not rely on AI to confirm a buyer’s qualifications. As for buyers, AI should not be trusted to make financing recommendations or determine someone’s overall budget. In terms of pricing, while AI can provide comps in the neighborhood, real estate professionals will always provide a deeper level of comp analysis. For instance, the caliber of renovation or the uniqueness of a particular property can only be sussed out by the human eye.”
AI can’t verify information
“AI can help write the words in our listing descriptions, but it can’t verify facts,” says Lisa Harris, with REMAX Center in Braselton, Georgia. “The top six areas I’ve seen most errors include Fair Housing comments, square footage, pricing, comps, schools and inaccurate marketing claims. I like to tell my clients that AI is like a blender, not a chef. It can mix info from various sources, but it takes a professional to analyze and compile the final product.”
